b Accumulation of Ad5 at the MTOC impairs Ad5 genome import

I. At the MTOC, CRM1 is required for genome delivery

I.2. b Accumulation of Ad5 at the MTOC impairs Ad5 genome import

We previously showed a blocking of Ad5 at the MTOC upon LMB treatment, leading to a defect in NPC translocation (Figure 14). Blocking of this step is expected to have an impact on downstream events, i.e nuclear import of Ad5 genomes. To confirm this, we analysed the effect of CRM1 inhibition by LMB treatment on Ad5 genome import. We performed Ad5 infections in the presence of LMB and quantified the number of nuclear pVII dots (Figure 19). No signal for pVII was detectable upon infection with LMB, even after 4 h pi. Moreover, capsids trapped at the MTOC were intact, since no pVII signal was detectable neither in the nucleus nor at the MTOC area.

Results

92 Figure 19. MTOC accumulation caused by LMB treatment leads to a defect in Ad5 genome import. U2OS cells were treated with (+ LMB) or without LMB (- LMB) for 45 min. Infections with Ad5-GFP particles were performed in the presence (+ LMB) or absence of LMB (- LMB) for 30 min up to 4 h. (A) Cells were fixed and stained with anti-Ad5 capsids (red) and anti-pVII (green) antibodies and with DAPI (grey) for chromatin staining. Images represent cells after 1 h of infection. Cells were imaged by confocal microscopy and maximal projection images are shown. (Scale bars, 20 µm). (B) Scatter plot showing the quantification of the total number of pVII foci colocalizing with DAPI signal per cell (depicted in (A): Merge DAPI/pVII), in the absence (black dots) or presence (red dots) of LMB.

Mean values (+/- SD) of 30 cells per condition are shown. Statistical analysis was performed using one-way ANOVA multicomparison test.

Results

93 I.2.c CRM1 is required for genome import and gene expression

Ad5 genome has to be delivered in the nucleus in order to initiate genome replication and expression. As previously showed, blocking of Ad5 capsids at the MTOC lead to a defect in Ad5 genome delivery. Thus, this MTOC retention inhibits nuclear steps of Ad5. Transcription of early genes is required to promote total transcription of Ad5 genes. 1 to 2 h pi, immediate early transcripts (E1A) can be detected (Berk et al. 1979; Glenn and Ricciardi 1988). However, up to now there is no available tool for the direct detection of Ad5 transcript via fluorescence microscopy.

We therefore developed and optimized a new protocol to visualize Ad5 E1A mRNA molecules at the single cell level. We adapted the RNAscope® Multiplex Fluorescent Assay (from ACDBio) in our infection model (see II.13 RNAscopeof Material and Methods section, for a detailed description). Cells were infected with Ad5 and fixed at different time points. Briefly, a probe mixture of 17 individual target sequences specifically designed to hybridize to Ad5 E1A mRNA transcripts was added for 2 h at 40 °C to the cells. Hybridized probes on the target were detected and signals were amplified with amplifiers in order to be detected by fluorescence microscopy. One fluorescent dot was considered as one E1A transcript, thus the total number of E1A transcription products can be quantified by fluorescence microscopy. Such assays can be combined with immunofluorescence staining, in order to visualize Ad5 capsids and genomes. As a specificity control, we infected U2OS with Ad5 vector lacking the coding E1A region (Figure 20). These cells were infected and Ad5 genomes correctly imported as Ad5 capsids and pVII signals were detectable. However, no signal for E1A mRNA was detectable.

When cells were infected with replicative Ad5 particles (comprising the E1A coding region), E1A dots were detected, after 2 h pi. This new method can be used to monitor gene expression overtime.

Results

94 Figure 20. Identification of Ad5 E1A mRNA by fluorescence microscopy is specific. U2OS cells were infected with Ad5-GFP vector deleted for the E1A region (top row) or Ad5 replicative particles (bottom row) for 2 h. Cells were fixed and E1A transcripts (magenta) were detected using specific RNA probes (RNAscope). A second staining using antibodies was used to detect Ad5 capsids (red) and pVII (green) and DAPI (grey) was used to stain chromatin. Cells were imaged by confocal microscopy and maximal projection images are shown. (Scale bars, 20 µm).

We then performed RNAscope assays upon LMB treatment (Figure 21). In the absence of LMB, nuclear E1A mRNA dots started to be detectable after 2 h pi. The number of E1A dots increased overtime, and 6 h pi E1A signals were mostly found in the cytoplasm. In comparison, MTOC accumulation of Ad5 capsids induced by LMB treatment led to impaired gene expression, as no E1A mRNA molecules were detected under these conditions.

Results

96 Figure 21. Functional CRM1 is required for Ad5 gene expression. (previous page) U2OS cells were infected with Ad5 replicative particles for 4 h (a and c) or 6 h (b and d) in the absence (-LMB) or presence (+ LMB) of LMB. Cells were fixed and E1A transcripts (magenta) were detected using specific RNA probes (RNAscope). A second staining using antibodies was used to detect Ad5 capsids (red) and pVII (green) and DAPI (grey) was used to stain chromatin. Cells were imaged by confocal microscopy and maximal projection images are shown. (Scale bars, 20 µm). (B) Scatter plot showing the quantification of the total number of E1A foci signal per cell in the absence (black dots) or presence (red dots) of LMB (pictures depicted in (A)). Mean values (+/- SD) of 30 cells per condition are shown.

Statistical analysis was performed using one-way ANOVA multicomparison test.

We confirmed with quantitative data that functional CRM1 is required for NPC translocation, leading to genome import and gene transcription. Upon LMB treatment, Ad5 genome was not detectable via pVII staining: core DNA is not exposed due to Ad5 capsid protection. We then focused our experiments to study the role of CRM1 in Ad5-genome release.

I.2.d CRM1 affects Ad5 capsid disassembly in mitotic cells

Several studies demonstrated a role for NPCs in capsid disassembly, e.g. via binding with Nup214 (Greber et al. 1997; Trotman et al. 2001; Strunze et al. 2011; Cassany et al. 2015). It was shown that the N-terminal part of Nup214 is required for the docking of Ad5, via the hexon protein, before genome release. These studies were performed in the context of intact NE, with assembled NPCs. In order to bypass this physical barrier of NE and to study if CRM1 possesses a role in capsid disassembly independent of the NPC, we established a protocol for Ad5 infection of mitotic cells (protocol established by Dr. I. Carlón Andrés, PhD thesis Irene Carlón-Andrés, 2017). In such a cellular model, every component of the NE and NPCs should be available in the cell, but not in the physiological context of an intact nucleus. Therefore, detection of pVII in mitotic cells is the result of direct capsid disassembly and not genome import because the NE barrier is absent.

Cells were synchronised in mitosis with colcemid (also known as demecolcine) treatment. This drug induces microtubules depolymerization and blocks cells in metaphase. Before Ad5 infection, cells were treated with or without LMB to analyse the role of CRM1 in Ad5 capsid disassembly. Infections of mitotic cells were done in colcemid-free medium and were analysed for up to 2 h pi, since after 2 h cells started to divide due to the reversibility of the colcemid block. pVII dots were detectable in mitotic infected cells 1 h pi, and increased overtime, resulting from capsid disassembly (Figure 22 A and B). Capsid disassembly can be observed in fixed cells by colocalization events between pVII and Ad5 capsids (Figure 22 A, upper row).

However, upon LMB treatment, no pVII were detectable, even at 2 h pi.

Results

97 Figure 22. Intact nuclear envelope is not required for Ad5 capsid disassembly. U2OS cells were treated with colcemid for 14 to 16 h to synchronise cells in mitosis. Cells were treated with (+ LMB) or without LMB (- LMB) for 45 min in the presence of colcemid. Synchronised cells were infected with Ad5-GFP particles with (+ LMB) or without LMB (-LMB) but in the absence of colcemid for 30 min up to 2 h. (A) Cells were fixed and stained with anti-Ad5 capsids (red) and anti-pVII (green) antibodies and with DAPI (grey) for chromatin staining. Colocalization events between Ad5 capsids and pVII are shown with white arrows. Cells were imaged by confocal microscopy and maximal projection images are shown. (Scale bars, 10 µm). (B) Scatter plot showing the quantification of the total number of pVII foci signal per cell in the absence (black dots) or presence (red dots) of LMB (pictures depicted in (A)). Mean values (+/- SD) of 30 cells per condition are shown. Statistical analysis was performed using one-way ANOVA multicomparison test.

Results

98 Our results showed that the addition of LMB impaired capsid disassembly and genome release in mitotic cells. In mitotic cells, there is no compartmentalisation between cytoplasmic and nuclear factors. Therefore, CRM1 cargoes blocked in the nucleus upon LMB treatment of interphase cells are found everywhere in mitotic cells and should be available for virus disassembly. Thus, it is unlikely that CRM1 cargoes sequestration in the nucleus induced by LMB is responsible for the disassembly defect. Our results strongly favour a direct role of CRM1 during capsid disassembly, where CRM1 dependent nuclear export is not required.

Moreover, an intact NE is also not required to perform Ad5 capsid disassembly.

I.2.e CRM1 promotes the total Ad5 genome release from the capsid

Our capsid disassembly analyses were based on antibody detection of pVII in fixed cells.

Fixation of cells can impair or hide some epitopes and the sensitivity of detection relies on the accessibility of these epitopes for antibodies. Moreover, single particle track analysis require live cell imaging experiments. To bypass these issues, our group had developed another indirect way of Ad5 genome detection, applicable to living cells (Komatsu et al. 2015). This system involves again pVII detection, but this time, via the detection of TAF-I. TAF-I is a cellular factor known to form ternary complexes with pVII on incoming genomes (Haruki et al. 2003).

Binding of TAF-I molecules to pVII upon genome exposure can then be monitored by fluorescence microscopy using U2OS TAF-I GFP expressing cell lines, generated in our lab by Dr. T. Komatsu.

Upon infection of these cells with Ad5, we clearly observed nuclear TAF-I GFP dots and all of them corresponded to pVII dots, as shown by the merge between TAF-I GFP and pVII channels (Figure 23, upper row). This system is specific, as TAF-I GFP dots were not detectable upon LMB treatment (Figure 23, lower row).

Results

99 Figure 23. TAF-I staining can be used for pVII detection. (previous page) U2OS cells stably transfected with a construct coding for TAF-I GFP were treated with (+ LMB) or without LMB (- LMB) for 45 min. Infection with Alexa 594 labelled Ad5-GFP particles was performed in the presence (+

LMB) or absence (- LMB) of LMB for 1 h. Cells were fixed and stained with anti-pVII (magenta) antibodies and with DAPI (grey) for chromatin staining. TAF-I was detected by GFP signal. Cells were imaged by confocal microscopy and maximal projection images are shown. (Scale bars, 20 µm).

We used TAF-I GFP U2OS cells in order to study the role of CRM1 in capsid disassembly in living cells. In this assay, the dynamic of capsid disassembly is resulting in pVII exposure and is monitored via the detection of TAF-I GFP dots overtime, by live cell imaging microscopy.

Cells were transfected with a construct coding for tagged Histone2B-tdiRFP to stain chromatin.

Cells were synchronised via colcemid treatment (as shown in section I.2.d CRM1 affects Ad5 capsid disassembly in mitotic cells), and infected with Alexa-594 labelled Ad5-GFP particles.

Infections were performed in the presence or absence of LMB. Mitotic cells were identified according to their chromatin staining (condensed chromosomes) and overall round shape.

Single cells were selected and followed overtime. Colocalization events between TAF-I and Ad5 capsid signals were considered as partial disassembled capsids. Under these conditions pVII (i.e Ad5 genome) was enough exposed to interact with TAF-I GFP, but the capsid remained partially intact to be detected via Alexa-594 labelling fluorophore. On the other hand, free TAF-I GFP dots were considered as completely released genomes, separated from capsids.

In non LMB treated control cells, approximatively 1 h 30 to 2 h pi, green TAF-I dots were detectable (Figure 24 A). Within the cell population several TAF-I dots were free from capsids (highlighted with filled white arrows), whereas some dots remained associated with capsids (highlighted with empty white arrows). The number of TAF-I dots free from capsid increased overtime (Figure 24 B). Moreover, virtually all free TAF-I dots and some TAF-I Ad5 associated dots were observed with a restricted mobility associated to cellular chromatin, implying that the genomes became stably anchored to the chromatin (zoom Figure 24 A).

In cells treated with LMB, TAF-I dots were also detectable. However, these TAF-I were exclusively associated with capsids (yellow dots, Figure 24 C), and the number of accumulating free TAF-I dots overtime was strongly decreased compared to control cells (Figure 24 D).

However, in the presence of LMB, chromatin targeting of partially disassembled capsids was also observed (zoom Figure 24 C).

Results

100

Results

101 Figure 24. Functional CRM1 is required for total Ad5 capsid disassembly in mitotic cells. (Fig A, B and C previous page) U2OS TAF-I GFP expressing cells were transfected with H2B-tdiRFP construct (blue) to stain chromatin. After 24 h of transfection, cells were treated with colcemid for 14 to16 h to be synchronised in mitosis. Cells were treated with or without LMB for 45 min in the presence of colcemid. Infection with Alexa-594 labelled Ad5-GFP particles was performed without colcemid but in the absence or presence of LMB. Mitotic cells were identified according to their chromatin staining (blue) and Ad5 capsids (red) as Ad5 genomes (TAF-I GFP dots; green) are depicted on the pictures.

Cells were imaged by spinning disk confocal microscopy. Maximal projection images are shown. (A) and (C) Mitotic U2OS TAF-I GFP cell treated without (A) or with (C) LMB. Maximal projection of signals detected in each channel at 130 min pi. TAF-I GFP dots free from Ad5 colocalization are shown with filled white arrows whereas TAF-I GFP dots colocalizing with Ad5 are shown with empty white arrows. (B) and (D) Overlay of TAF-I GFP (green) and Ad5 capsids (red) signals in one single cell in absence (B) or presence (D) of LMB overtime. From the top left corner (120 min) to the bottom right (137 min) each frame is separated by 1 min. TAF-I GFP dots free from Ad5 colocalization are shown with white arrows.

These results showed that capsid disassembly and genome separation in mitotic cells require functional CRM1. The strong reduction of free TAF-I dots observed upon LMB treatment suggests that inhibition of CRM1 impairs Ad5 genome capsid-release. In contrast, in mitotic cells, partially disassembled capsids were targeted to the chromatin, even in the presence of LMB. One hypothesis is that a partially exposed core-genome is sufficient to target the genome to chromatin, dragging the attached capsid with it. Complete genome release and capsid disassembly, however, would need functional CRM1. In fixed mitotic cells, antibody detection of pVII in LMB treated cells did not give any signal (Figure 22), whereas pVII could be detected using the TAF-I GFP system. The TAF-I GFP pVII detection system appears thus more sensitive and does not rely on epitope recognition. However, we have not tested pVII detection in TAF-I GFP U2OS mitotic fixed cells.

Results

102 Our analyses in U2OS cells confirmed previous studies about the role of CRM1 in efficient nuclear genome import. During the first steps of infection, neither CRM1 nor other nuclear factors are required for Ad5 trafficking to the MTOC. Ad5-MTOC interaction is not well characterized but our data are in favour of an interaction independent of the integrity of the microtubule network. However, functional CRM1 is needed to mediate Ad5-MTOC removal for NPC translocation. Inhibition of CRM1 with LMB impairs nuclear genome import, leading to a defect in Ad5 gene expression. Our model of mitotic infected cells gave us more insights into the role of CRM1. Total genome release from Ad5 capsid requires functional CRM1 and it seems to directly involve CRM1 and none of its cargoes.

II. A new CRM1 mutant as a tool to study Ad5 genome import

CRM1, the major cellular export factor, is known to form a ternary complex together with RanGTP and NES-containing cargoes (Ossareh-Nazari and Dargemont 1999; Fornerod et al.

1997b; Monecke et al. 2013). Its final binding site on Nup214 has been shown to promote the efficient release of some export complexes (Kehlenbach et al. 1999; Bernad et al. 2006; Hutten and Kehlenbach 2006). Moreover, our group showed that CRM1 binds some FG-repeats of Nup214 (Roloff et al. 2013) and solved the crystal structure of the export complex CRM1-RanGTP-SPN1 associated with FG-repeats fragment of Nup214 (Port et al. 2015). Based on this study, several mutants of CRM1 were generated. In order to study CRM1 and its interacting partners upon Ad5 infection, we tested a batch of these mutants in our infection assays. Interestingly, one of these mutants did not show any defect in Nup214 binding assays (data not shown) but was found to impair Ad5 genome delivery. We chose to study in details this CRM1 mutant in order to better characterize its role upon Ad5 infection. We generated and characterized new cell lines constitutively expressing the CRM1 mutant. We also performed export assays and biochemical studies with recombinant proteins to characterize the export kinetic and the binding with NES in this mutant, in comparison to the wild type protein.

II.1 CRM1 W142A P143A mutation

A CRM1 mutant library was initially generated by Dr. S. A. Port, to study the interaction between CRM1 and Nup214-FG repeats. Based on the predicted structure of CRM1 in complex with RanGTP and SPN1, bound to an FG-repeats fragment of Nup214, point mutations were introduced on CRM1 at the predicted binding site of Nup214. In export complexes, RanGTP is found in the central domain of CRM1 whereas the NES cargo interacts with the outer surface of CRM1 (reviewed in (Monecke et al.2014)).

Results

103 Therefore, as an example for one of the CRM1 generated mutants, W142 P143A mutations are located close to the site of interaction between CRM1 and Nup214 (but not overlapping) (Figure 25).

Figure 25. Mutations W142A P143A are close to the binding region of Nup214 FG-repeats.

Overall structure of CRM1 (blue) in complex with RanGTP (light orange), the cargo SPN1 (cyan) and Nup214-FG repeats (red). Different orientations of the complex are depicted. Mutations W142A P143A on CRM1 are highlighted in yellow (Port et al. 2015).

In order to study the impact of a mutation on CRM1 function in the cell, endogenous CRM1 has to be inactivated with LMB treatment. In this context, all mutants tested had to be LMB resistant. Therefore, the C528S mutation was introduced in addition to other mutated sites in every CRM1-mutant tested. A preliminary screening of several CRM1 mutants from the mutant library mentioned above was done by Dr. I. Carlón-Andrés. U2OS cells were transfected with various HA-tagged and LMB-resistant CRM1 and infected with Ad5. The efficiency of capsid disassembly in mitotic cells was quantified according to the number of pVII dots normalised to the total number of Ad5 capsids per cells, after 1 h of infection, upon LMB treatment (Figure 26, adapted from PhD thesis Irene Carlón-Andrés, 2017).CRM1 C528S (conferring the LMB resistance) increased the capsid disassembly efficiency, compared to cells non-treated with LMB, confirming a role of CRM1 during this process. Five other mutants holding 2 (or 4) extra mutations were able to rescue capsid disassembly upon LMB treatment. Only CRM1 with W142A P143A mutations was not able to rescue the LMB inhibition effect on capsid disassembly.

Results

104 Figure 26. CRM1 W142A P143A C528S is not able to rescue Ad5 capsid disassembly upon LMB treatment. U2OS cells were transfected with empty plasmid (mock) or with CRM1 C528S-HA constructs with specific mutations, indicated below the X axis. Cells were treated with colcemid for 14 to 16 h to synchronise cells in mitosis and treated with LMB (red, grey and blue conditions) or without (black condition) for 45 min prior to infection. Synchronised cells were infected for 1 h with Alexa 594 labelled Ad5-GFP particles with or without LMB but in absence of colcemid. Cells were fixed and stained with anti-HA antibodies to identify transfected cells. Ad5 genomes were identified via anti-pVII antibodies and DAPI was used for chromatin staining. Cells were imaged by confocal microscopy and the quantification of the number of pVII foci normalised to the number of Ad5 capsid per cell was performed on maximal projection images. Results are depicted with a scatter plot. Mean values (+/- SD) of 30 cells per condition are shown. Statistical analysis was performed using one-way ANOVA multicomparison test, comparing every condition to mock transfected LMB treated U2OS cells (red condition on the graph) (modified from PhD thesis Irene Carlón-Andrés, 2017).

II.2 Generation and characterization of CRM1 mutant expressing cell lines

To further analyse the Ad5 capsid disassembly defect observed upon expression of the CRM1 W142A P143A C528S mutant, we generated U2OS cells constitutively expressing this mutant form of CRM1. The CRM1-HA construct was transfected in U2OS and after few days of culture, LMB was added to the medium, to select cells that had incorporated the LMB resistant form of CRM1. Indeed, random events of integration can be observed upon transfection of mammalian

Im Dokument From the centrosome to the nuclear envelope and beyond: insights into the role of CRM1 in adenoviral genome delivery (Seite 96-0)